Fig. 2.

The rePrime procedure is demonstrated for the molecule 2-hydroxy isophthalate (2hipa). a The moiety size (λ) indicates the distance between nodes in a molecular graph. At λ = 1, the moiety at node n (the carbon atom shaded in orange) is the atom itself. The moiety centered at node n, the number of unique moieties and reaction rules extracted from MetRxn at $\lambda =\left\{1,2,3\right\}$ are provided on the table. b, c The rePrime procedure iteratively assigns prime number to each node at λ = 1, 2 (see Supplementary Methods for details). The molecular signature of 2hipa, $C_{m,\mathrm{2hipa}}^{\lambda }$ contains the count of unique moieties (i.e., prime number). The resulting $C_{m,\mathrm{2hipa}}^{\lambda }$ increases in specificity at larger moiety size (λ) as neighboring atoms are also involved in determining the canonical label of the node. d The derivation of reaction rule for two decarboxylase reactions (i) 2-hydroxyisophtalate decarboxylase (2HIPD) and (ii) salicylate decarboxylase (SLD) are demonstrated at λ = 1. The resulting reaction rules $T_{m,\mathrm{2HIPD}}^1$ and $T_{m,\mathrm{SLD}}^1$ are identical at this moiety size, hence they are stored as $T_{m,1}^1$. The specificity of the reaction rules increases with increasing moiety size λ, therefore $T_{m,\mathrm{2HIPD}}^2$ is not equal to $T_{m,\mathrm{SLD}}^2$ (Supplementary Table 3)